intelligent machines and the networked world – artificial intelligence
Artificial intelligence (Al) is a specialized branch of computer science. All researchers attempt to imitate some of the abilities of living things, such as perception, learning, logical reasoning, and language using computers.
Individual biological nerve cells (neurons) can be simulated by computers and connected to each other in layers, forming a so-called neural network. Data entering the input layer is processed by the network and passed along to the output layer. The data processed in the outer layer depends on the pattern of connections between the individual neurons.
During the learning phase, the connections are adjusted in accordance with sample data, so that after- wards the system can identify such items as handwritten letters, even if each letter is not always written in the same way.
Computer-based knowledge
Storing knowledge is a complicated process and provides a much greater challenge than simply recording ordinary data. Knowledge about a chair, for instance, goes far beyond simple information about its shape and location. In many contexts its function as a place to sit may be consider- ably more important. Beyond this basic function, a chair can also be used, for example, as a surface to rest objects or as a stepstool.
There is also knowledge about events and actions, and even knowledge about knowledge itself-such as its scope, reliability, and origin. This “knowledge data” can be combined into statements. By using several statements, logical rules can be used to derive further statements.
These methods are used by so-called expert systems to figure out, for instance, what has caused a machine to malfunction and even to plan its repair.
Al and robotics
The first robots were programmed according to the model “perceive => plan => act.” The downside of this technique is the great deal of computational time required, and the slow reaction speed of the machines when unexpected events arise.
Later, researchers imitated the stimulus-response principle followed by living things, such as that used when we reflexively pull our fingers away from a hot stove. These complicated stimulus-response chains are used, for example, when we are climbing stairs while being simultaneously absorbed in a conversation.
Today, both methods are combined for programming robots to complete complex missions.
THE TURING TEST
about knowledge itself-such as its scope, reliability, and origin. This “knowledge data” can be combined into statements. By using several statements, logical rules can be Later, researchers imitated the stimulus-response principle followed by living things, such as that used when we reflexively pull our fingers away from a hot stove.
These complicated stimulus-response chains are used, for example, when we are climbing stairs while being simultaneously absorbed in a conversation. Today, both methods are combined for programming robots to complete complex missions. In the 1950s, mathematician Alan Turing devised a test to clarify the question of whether a computer can be
regarded as intelligent.
The structure of the test requires a human judge to engage in a natural language conversation on a computer. They must identify whether they are conversing with a human or a computer programmed with human-like responses. If the computer can be identified, it has failed the Turing test.